Current quality source-adaptive schemes for multicast multilayered video rely on merging capabilities at special nodes in the network as a means of combining feedback from the whole set of receivers. These strategies reduce network load and avoid feedback implosion at the source. In this paper, we examine how to provide optimal feedback for such schemes. The optimal feedback is achieved when state information from the whole set of receivers is represented in every incoming feedback packet at the source. We show that the choice of a suitable merging time window in the intermediate nodes coupled with a periodical transmission of feedback packets by the receivers leads to near-optimal feedback.